Study on the method and system for falling detection based on the inclinometer / 生物医学工程学杂志

A falling is a risky incident of safety and health of human. It may cause serious injuries, such as bone fracture, and even death. A falling detection method based on inclinometer is described. At first, we collect angle data recorded by a wearable inclinometer placed at subject's waist. The angular data are transmitted to PC through a wireless data transmission device. Then, the falling duration is divided into three phases: the state of fall, the impact phase, and the posture phase. We make threshold-based fall-detection decisions in every phase after feature extraction and analysis of the short-time angle data. Finally, a robust falling detection result is given by comprehensive considerations of the three phases decisions. The experiment results proved that the accuracy of our falling detection method was up to 97.23% without undetected falls.

An FES system based on dual axis inclinometer in foot drop treatment / 生物医学工程学杂志

The foot drop functional electrical stimulation (FES) system consisting of various sensors has been widely applied to the disease of the foot drop. However, the current system is limited to the research on walking on the ground and ignores other important actions of foot in one's daily life, such as walking up and down the stairs, squatting and lying down, etc. In this work, we applied the dual axis angle sensor to the system of the foot drop FES for the first time. Such a system can not only stimulate the foot drop during normal walking, but also identify squatting, sitting, and lying down etc. and furthermore, the system can switch off automatically. In the meanwhile, it can also detect falls and other dangerous actions. The accuracy of our system can achieve 100%, 81.9%, 95.8%, 99.0% and 66.9% for normal walking, sitting-standing, walking up the stairs, walking down the stairs and squatting-standing respectively.

The analysis method of the Hilbert spectrum entropy of dividing frequency range for signals of heart rate variability / 生物医学工程学杂志

The signal analysis of heart rate variability (HRV) has been very significant for heart disease of aided diagnosis, monitoring and evaluation. We proposed a new method of HRV signal analysis based on the Hilbert spectrum entropy dividing frequency range. According to Hilbert spectrum characteristics of the multi-resolution and the characteristic of HRV signal frequency spectrum, the Hilbert time-frequency spectrum entropy of HRV signal in different frequency range and the full frequency Hilbert time-frequency spectrum entropy with weighting factor were calculated. This approach was analyzed after the appropriate separation for various physiological factors based on the frequency range and it is more conducive to reflect the physiological and the pathological characteristics. Applying the new approach to the actual HRV signal of the MIT-BIH standard database, we obtained the results which showed that this method could effectively differentiate from the sample group for the young, the elder and the patients with atrial fibrillation, and for the sample group for the healthy persons and CHF patients, the performance in statistical analysis was superior to those of the general time-frequency entropy method. The approach could provide an effective analysis method for clinical HRV signal.

Estimation of evoked potentials based on MD criterion and Givens matrix in non-Gaussian noise environments / 生物医学工程学杂志

Traditional EP analysis is developed under the condition that the background noises in EP are Gaussian distributed. Alpha stable distribution, a generalization of Gaussian, is better for modeling impulsive noises than Gaussian distribution in biomedical signal processing. Conventional blind separation and estimation method of evoked potentials is based on second order statistics (SOS). In this paper, we propose a new algorithm based on minimum dispersion criterion and Givens matrix. The simulation experiments show that the proposed new algorithm is more robust than the conventional algorithm.

Texture analysis of SPIO-enhanced MR imaging in rat models of hepatocellular carcinoma and hepatocirrhosis based on gray level co-occurrence matrix / 中国医学影像技术

Objective To analyze the texture features of SPIO-enhanced MR imaging in rat models of hepatocellular carcinoma (HCC) and hepatocirrhosis with gray level co-occurrence matrix (GLCM). Methods HCC and hepatocirrhosis models were established in rats. SPIO-enhanced MR images were obtained. A total of 161 regions of interests (ROIs, 81 of HCC and 80 of hepatocirrhosis) were selected manually. Feature values as angular second moment, contrast, correlation, inverse difference moment, entropy, variance were extracted based on GLCM. The differences of feature values between two groups were statistically analyzed. Results In SPIO-enhanced MR images, hypointense signal changes were found in hepatocirrhosis, as well as hyperintensity in HCC nodules and intermixed intensity in larger HCC nodules. Correlation and entropy values of HCC group were higher than that of hepatocirrhosis group, while the angular second moment, contrast, inverse difference moment, and variance values were lower than hepatocirrhosis group. Conclusion The feature values based on GLCM could be used for the further computer aided diagnosis of SPIO-enhanced MR images in rat models of HCC and hepatocirrhosis.

Technological progress of computer-aided detection and diagnosis of lung nodule based on CT image analysis / 国际生物医学工程杂志

Lung nodules are one of the most common pathological changes, thus early detection of lung nodule is very important for the diagnosis medical treatment of lung eancer. In recent years, as the application of multi-slice spiral CT(MSCT), high-resolution CT(HRCT) and low-dose chest CTCLDCT), computer-aided diagnosis (CAD) system will be more essential and more important. Since CAD system can improve the working efficiency of doctors and provide service to more patients, has become the research hotspot and achievement has been made in relevant area internationally recently. This review summarizes the basic methods and applieations of computer-aided detection and diagnosis of lung nodule based on CT image.

A novel method to quantitatively analyze the fatty liver ultrasonic images based on multiresolution fractal Brownian motion model and genetic algorithm / 生物医学工程学杂志

Medical ultrasonic imaging is frequently used for diagnosing the fatty liver disease. In order to help doctors diagnose fatty liver disease more precisely, we need to construct a quantitative assessment system, and in this paper, we propose a method to construct such system with the use of multiresolution fractal Brownian motion model and the genetic algorithm. In such a way, a set of standards can help doctors diagnose the degree of the fatty liver disease more precisely.

An EMD based epileptic spike detection method / 生物医学工程学杂志

The automatic spike detection in EEG is significant in both diagnosing illness and alleviating the heavy labour force of the doctor. This paper proposes a new EMD based method to complete spike detection. It decomposes a signal into a few intrinsic mode functions (IMF), and then applies the nonlinear energy operator (NEO) to the first IMF to complete the automatic detection. Sufficient results are obtained by applying this method to the spike detection of the simulation signal and the real epileptic EEG signal.

Research on chaotic behavior of epilepsy electroencephalogram of children based on independent component analysis algorithm / 生物医学工程学杂志

In this paper, Independent component analysis (ICA) was first adopted to isolate the epileptiform signals from the background Electroencephalogram (EEG) signals. Then, by using the phase space reconstruct techniques from a time series and the quantitative criterions and rules of system chaos, different phases of the epileptiform signals were analyzed and calculated. Through the comparative research with the analyses of the phase plots, the power spectra, the computation of the correlation dimensions and the Lyapunov exponents of the physiologyical and the epileptiform signals, the following conclusions were drawn: (1) The phase plots, the power spectra, the correlation dimensions and the Lyapunov exponents of the EEG independent components reflect the general dynamical characteristics of brains, which can be taken as a quantitative index to weigh the healthy states of brains. (2) Under normal physiological conditions, the EEG signals are chaotic, while under epilepsy conditions the signals approach regularity.

Spike extraction of epileptic waves in EEG based on EMD / 生物医学工程学杂志

The automatic spike detection in EEG is significant in both diagnosing epilepsy and alleviating the heavy labor force of the doctors. This paper proposes an empirical model decomposition (EMD) based epileptic spike detection method. It extracts the high frequency components related to spikes in EEG signal by EMD, and it detects the spikes by calculating the instantaneous amplitude of the high component with Hilbert transform. The results of experiments show that the method works well.

An EMD based time-frequency distribution and its application in EEG analysis / 生物医学工程学杂志

Hilbert-Huang transform (HHT) is a new time-frequency analytic method to analyze the nonlinear and the non-stationary signals. The key step of this method is the empirical mode decomposition (EMD), with which any complicated signal can be decomposed into a finite and small number of intrinsic mode functions (IMF). In this paper, a new EMD based method for suppressing the cross-term of Wigner-Ville distribution (WVD) is developed and is applied to analyze the epileptic EEG signals. The simulation data and analysis results show that the new method suppresses the cross-term of the WVD effectively with an excellent resolution.

The sample entropy and its application in EEG based epilepsy detection / 生物医学工程学杂志

It is of great importance for the detection of epilepsy in clinical applications. Based on the limitations of the common used approximate entropy (ApEn) in the epilepsy detection, this paper analyzes epileptic EEG signals with the sample entropy (SampEn) approach, a new method for signal analysis with much higher precision than that of the ApEn. Data analysis results show that the values from both ApEn and SampEn decrease significantly when the epilepsy is burst. Furthermore, the SampEn is more sensitive to EEG changes caused by the epilepsy, about 15%-20% higher than the results of the ApEn.

A Blind Source Separation Algorithm Based on Beamform Theory / 航天医学与医学工程

Objective To deal with blind source separation(BSS) more effectively in the field of mixed signal separations of strong and week sources.Methods According to the consistency between array signal processing model and BSS model,the real sources were estimated under linear constrains and least mean square(LMS),based on minimum output energy(MOE).EEG and evoked potential(EP) were used as strong background noise and week signal source separately in our experiment.The mixed signals were separated with the method proposed in this paper.Results The EP could be seperated from the strong noise EEG effectively.Conclusion Compared with typical BSS approaches,this new algorithm need not solve the unmixing matrix,so it runs fast,is of a little low computational complexity and can correctly estimate the weak signal source from low signal/noise(S/N) ratio.

The development of algorithms for adaptive latency change detection of evoked potentials under alpha-stable noise conditions / 生物医学工程学杂志

The latency change detection of EPs is of special interest in many clinical applications such as diagnosis of the injury and pathological changes in the nervous system. This paper reviews the adaptive latency change detection approaches under stable noise conditions, based on the fractional lower order statistics. It also evaluates and compares the performances of the presented algorithms.

Progress in automatic detection of epilepsy based on EEG analysis / 生物医学工程学杂志

Automatic detection of epileptic events is of significance in clinical application. It is helpful to reduce the electroencephalogram analysts' workload. This paper summarizes and analyzes the detection of epileptic events by traditional and especially advanced methods, including nonlinear filtering, template matching, mimetic, wavelet transform and artificial neural network.

The algorithms and development for the extraction of evoked potentials / 生物医学工程学杂志

The extraction of evoked potentials is a main subject in the area of brain signal processing. In recent years, the single-trial extraction of evoked potentials has been focused on by many studies. In this paper, the approaches based on the wavelet transform, the neural network, the high order acumulants and the independent component analysis are briefly reviewed.

Knowledge discovery in database and its application in clinical diagnosis / 生物医学工程学杂志

Nowadays the tremendous amount of data has far exceeded our human ability for comprehension, and this has been particularly true for the medical database. However, traditional statistical techniques are no longer adequate for analyzing this vast collection of data. Knowledge discovery in database and data mining play an important role in analyzing data and uncovering important data patterns. This paper briefly presents the concepts of knowledge discovery in database and data mining, then describes the rough set theory, and gives some examples based on rough set.